Research Information System
Journal of Energy Storage, vol.152, 2026 (SCI-Expanded, Scopus)
Latent heat storage systems are hindered by the low thermal conductivity of phase change materials (PCMs) and the limited adaptability of current designs. In this study, a novel triangular-rib enhanced heat storage system was developed. To evaluate system performance, two artificial neural network (ANN) models were trained to predict the melting time at 50% and 100% liquid fractions as functions of tube diameter, system orientation, and orifice size. The ANN models exhibited high predictive accuracy, with R2 values of 0.996 (50% melting) and 0.992 (complete melting), RMSE values of 69.55 s and 20.37 s, and MAE values of 42.86 s and 15.93 s, respectively. Coupling the ANN outputs with a genetic algorithm enabled single- and multi-objective optimizations, yielding two optimal configurations. The ribbed designs significantly outperformed the core design, achieving complete melting in less than 5 h, while the core design reached only about 48% liquid fraction in the same period. These designs also enhanced stored energy by approximately 75% compared to the core design. A cost analysis further showed that the payback period decreased significantly from 972 days for the core design to about 842 days for optimized configurations, corresponding to an improvement of 13.4%.